Abstract
Wind energy can reduce reliance on fossil fuels, but identifying suitable sites requires balancing techno-economic feasibility with environmental and social constraints. Brazil has outstanding wind resources, yet environmental, social and land-use conflicts can limit deployment. Here, we develop a national-scale GIS-based multi-criteria decision-making framework for onshore wind farm site selection in Brazil, integrating 14 exclusion constraints and six evaluation criteria derived from the literature and national datasets. Criterion weights were elicited from Brazilian experts using the analytic hierarchy process. Under the Wind Power Expansion (WPE) scenario, 1.67 million km² (approximately 20% of Brazil) was classified as suitable, corresponding to a theoretical capacity of 15,812 GW. Under the more restrictive High Economic Reward (HER) scenario, 14,268 km² remained suitable across the mapped suitability classes. Suitable areas concentrate along the eastern coast and in the southern region. At 100 m spatial resolution, the resulting municipality-level rankings can support strategic planning, stakeholder engagement, and the sustainable expansion of onshore wind energy in Brazil.
Keywords
Highlights
Developed a GIS-based MCDM-AHP framework for national-scale onshore wind farm site selection in Brazil.
Calibrated exclusion buffers using existing wind turbine locations and literature benchmarks.
Identified ~20% of Brazil as suitable under the WPE scenario, with a theoretical capacity of 15,812 GW.
Mapped priority regions and ranked municipalities by the extent of suitable land for wind energy development.
Introduction
The urgent need to bridge the energy demand-supply gap with clean and renewable power has stimulated government policies and legislation worldwide to encourage investment in renewable energy (RE) in recent years. As environmental awareness and affordable renewable energy solutions continue to spread, wind energy has become one of the fastest-growing RE options in terms of clean fuel sources, job opportunities, and environmental sustainability (Adeyeye et al., 2020; Banerjee, 2022). Wind energy is the second-largest source of RE production globally after hydropower, due to its mature technology and extensive global supply chain (IEA, 2024). In the past two decades wind energy represented more than a 23-fold increase (IRENA, 2025), reaching 1133 GW by the end of 2024 in terms of global cumulative installed wind capacity (IRENA, 2025). In 2023, wind power has accounted for 7.8% of global wind electricity production (IEA, 2024). In this context, Brazil has emerged as one of the leading markets in the RE sector. Wind energy contributed 2.85% to Brazil’s energy consumption, with its new installed capacity reaching 30.45 GW in 2023 (ABEEólica, 2024). Despite 51% of Brazil’s energy demand still being met by non-renewable sources (EPE, 2024), the country exhibits substantial potential for further wind energy development. A recent report suggests that onshore wind capacity could meet twice the total energy demand projected for 2050 under the most expansive scenario, driven by significant electrification efforts (EPE, 2020). However, enabling this energy transition in Brazil requires planning, implementation strategies, and detection of good potential locations and selection of the optimal wind farm sites. These are the primary steps to harness sustainable energy resources (Seyed Alavi et al., 2022).
Wind power site selection is a complex process which must be evaluated from multiple perspectives, including techno-economic, social, and environmental ones (Asadi et al., 2023), and encompass factors such as visual impacts, noise generation, wildlife impact, and electromagnetic interference (Teff-Seker et al., 2022). For instance, economic feasibility is enhanced in sites with high mean wind speed and reasonable proximity to power transmission line and transportation infrastructure to minimize wind farm installation and maintenance costs (Rediske et al., 2021). Multi Criteria Decision Making (MCDM) methods are available to account for the complexity of decision making under imprecise and ambiguous conditions, particularly in the renewable energy field (Lak Kamari et al., 2020). To enable the organization, storage, manipulation, analysis and modeling of large amounts of data from the real world linked to a spatial reference grid, Geographic Information System (GIS) can be used in combination with such MCDM approach. It facilitates the incorporation of social, cultural, economic and environmental criteria that leads to a more effective decision-making (Rediske et al., 2021), and therefore provides a useful tool for selecting sites for wind farms (Villacreses et al., 2017).
GIS-based MCDM approaches for wind farm site selection have been extensively researched in recent years, applied in contexts such as France (Ifkirne et al., 2022), Morocco (Taoufik and Fekri, 2021), Turkey (Atici et al., 2015; Aydin et al., 2010; Daneshvar Rouyendegh et al., 2018), China (Li et al., 2020), the U.S. (Ajanaku et al., 2022; von Krauland et al., 2021), India (von Krauland and Jacobson, 2024), Saudi Arabia (Baseer et al., 2017), Iran (Bina et al., 2018; Moradi et al., 2020; Mostafaeipour et al., 2019; Yousefi et al., 2022), the Netherlands (Klok et al., 2023), and Ethiopia (Yegizaw and Mengistu, 2023). von Krauland and Jacobson (2024) and von Krauland et al. (2021) focused on land use restrictions and altitude adjustments with generalized wind speed thresholds. These studies typically apply a similar framework and model, and also consider a set of constraint factors and criteria that vary based on the study area profile and research objectives (Table 1). It was noticed that most studies set their minimum safety distances from constraint factors based on established literature, without adapting or updating them to local conditions of the study area. Furthermore, there are a limited number of studies focusing on Latin America. Specifically, Archer and Jacobson (2005) assessed global wind potential, including Brazil, using reanalysis data; Villacreses et al. (2017) explored onshore territories in Ecuador; Weiss et al. (2018) studied three neighboring municipalities in Southern Brazil; and Vinhoza and Schaeffer (2021) conducted the first nationwide assessment of offshore wind farm suitability along Brazil’s coast. No nationwide onshore wind farm site selection study for Brazil has been conducted before.
Summary of selected GIS-based multi-criteria decision-making studies on wind farm site selection (the last three studies focus on Latin America).
Occupational repetitive actions.
Technique for order of preference by similarity to ideal solution.
Furthermore, the environmental impacts and social injustices associated with wind farm projects have raised considerable concerns across Brazil. A recent literature review study evaluated 25 studies focused on Northeast Brazil (da Silva and Galvão, 2022), where over 87% of the country’s wind farms are installed. It revealed various social issues including road closures, house cracking, land privatization, and loss of land tenure affecting local communities, along with environmental impacts such as dune destruction, pond backfilling, coastal erosion, and reduction of fauna and flora. Similar findings were reported in other studies carried out in this region (Brannstrom et al., 2017; Gorayeb et al., 2018; Nascimento, 2019). While studies focused on the South of Brazil are less common, the connection between wind farm projects there and negative impacts on vulnerable biomes (Turkovska et al., 2021) and the perpetuation of social disparities has also been shown (Lenoir-Improta and Di Masso, 2021).
This is the first nationwide onshore GIS based-MCDM wind farm site suitability study for Brazil. Based on the abovementioned complex background, we incorporated a diversity of Brazil-specific constraints relevant to wind farm site selection, as a mean to support an effectively sustainable green energy transition in the country. The model we developed is calibrated using statistical analysis of existing wind turbine locations combined with values from literature, which enhances the accuracy of site suitability in Brazil’s socio-environmental context. Two scenarios were assessed: one to identify the potential of wind energy expansion in Brazil, and another one to highlight the most promising region for wind energy exploitation. Based on this previous one, top Brazil municipalities were ranked by extent of suitable area.
Methodology and materials
This study exclusively examined onshore wind farm suitability. This decision was based on the fact that Brazil still has plenty of space for onshore wind farm development and that the assessment for offshore potential demands a new methodology, with different set of criteria. Brazil’s continental territory spans approximately 8.51 million km² (Supplemental Table S1), with its 26 states and the Federal District categorized into five regions: North, Northeast, Southeast, South, and Central-West (Supplemental Figure S1).
Methodology framework
Our methodology framework of wind farm site selection presents four main steps (Figure 1): (1) Identification of constraint factors and evaluation criteria for wind farm site selection (Table 2); (2) Definition of the minimum distance from future onshore wind farms to constraint factors, which is calibrated based on the statistical analysis of existing wind turbines and values from literature (Supplemental Table S2); (3) Determination of evaluation criteria and assessment of each criterion’s weights using the Analytical Hierarchy Process (AHP) method (Table 3); and (4) Integration of constraint factors map and evaluation criteria map using the Simple Additive Weighting Method (SAW) to generate a wind farm suitability map.
Methodological framework and workflow of the present study.
Constraint factors and exclusion buffers considered in the present study.
A minimum distance of 750 m from existing wind turbines was adopted in this study (Gupta, 2016), to mitigate wake effects that reduce downstream turbine efficiency, balancing this with the benefits of leveraging existing infrastructure.
500 m of minimum distance was applied to the main rivers, while 200 m was considered for all the other water bodies, including minor rivers, lakes, and others.
Analytic hierarchy process (AHP) pairwise comparison matrix and resulting weights for the evaluation criteria.
Identification, calibration, and integration of constraint factors
Identification of constraint factors
A total of 14 constraint factors were selected based on a review of wind farm site selection literature (Table 1) and the environmental and social issues specific to Brazilian territory. The 14 factors are described below and are categorized into three groups: technical, environmental, and social. These factors and the related data to subsidize the definition of their respective minimum distances are given in Supplemental Table S2.
Urban settlements
Most of the consulted studies have identified urban areas as a significant constraint for wind farm site selection (Ayodele et al., 2018; Baseer et al., 2017; Ifkirne et al., 2022; Klok et al., 2023; Moradi et al., 2020; Villacreses et al., 2017; Yegizaw and Mengistu, 2023). Urban settlement data (Brasil, 2022) for Brazil, derived from a 2019 national satellite imagery survey at 10 m resolution. This 2019 layer is the most recent nationwide urban settlement dataset publicly available from IBGE at the time of the analysis. It encompasses a variety of urban configurations, including densely populated urban centers, isolated urban settlements, rural villages, and isolated housing complexes with more than 10 buildings or 50 residences within a 50-meter radius (IBGE, 2022). It is desirable to maintain a minimum distance from operational wind farms to urban settlements to mitigate nuisances such as noise, shading, and vibration affecting civilians (Rediske et al., 2021). However, longer distances between wind farms and larger urbanized areas and other high electricity consumption centers would increase electricity loss through transmission lines.
Environmental protection areas
These areas are legally categorized into two types in Brazil: “Strict Protection Areas,” which prohibits any natural resource exploitation to preserve ecological integrity, and “Sustainable Use Areas,” in which limited resource exploitation is permitted. This study regards all categories of environmental protection areas as restricted regions for wind farm construction to ensure the environmental adequacy of future projects. This is geospatial database updated in August 2019 (Brasil, 2021b), and encompasses 1805 Environmental Protection Areas spreading across various biomes, which sometimes overlap (Brasil, 2021b).
Water bodies and streams
Streams and their adjacent riparian vegetation are classified as areas of permanent protection under the native vegetation protection law (Taniwaki et al., 2018), with required buffer distance ranging from 10 to 500 m, depending on the type and width of the water resource (Brasil, 2012). The geospatial data used in this study does not specify the width of water bodies.
Indigenous territory
Brazilian government has recognized 690 territories for its indigenous population, covering about 13% of Brazil’s land mass (Survival International, 2025). These territories are classified as protected areas to preserve the cultural traditions and natural environments of these communities. Therefore, wind farm construction and operation should keep away from these territories, even though none of the reviewed studies have considered this constraint factor. The updated data (October 2023) for indigenous territories comprises 736 polygons, primarily concentrated in the Amazon Forest Area (Brasil, 2021a), that is considered as restricted area for future wind farm projects in this study.
Transportation infrastructure
This study utilizes a comprehensive dataset encompassing all national and state roads, ports, public airports, and the national railway system. Wind farms must be sited outside the established safety distances from these infrastructures, particularly from ports and airports, to avoid interference with radar systems. However, proximity to transportation infrastructure can significantly reduce construction and maintenance costs for wind farms. The geospatial data for transportation was updated in June 2023 (Brasil, 2023).
Important bird areas (IBAs)
Areas identified as IBAs are critical for birdlife and, consequently, for general biodiversity and encompass sites for bird breeding, resting, and habitat preservation, especially for endangered species. The data used in this study comprises 237 IBAs, covering approximately 11% of Brazil’s territory. While most of these areas are located within the Amazon Forest region, several are also distributed across other biomes within the country. Acknowledging the ecological significance of IBAs, they are considered as a constraint factor for wind farms in the present study (De Luca et al., 2009).
Altitude and slope
Despite a few notable peaks, including one reaching 2993 m above sea level, around 95% of Brazil’s terrain sits below 600 m in elevation. Furthermore, land slopes generally remain below 20%. Regions with higher elevations and steeper slopes can present challenges for installation and maintenance accessibility. Prior studies have identified areas with slopes exceeding 15% as unsuitable for wind farm construction (Ayodele et al., 2018; Ifkirne et al., 2022; Villacreses et al., 2017).
Wind speed
Mean wind speed at 100 m height was used in this study (DTU Wind Energy and World Bank Group, 2018), aligning with the average hub height of installed wind turbines in Brazil, approximately 112 m (EPE, 2020).
Existing wind turbines
Keeping a safe distance from existing wind turbines is necessary to mitigate the wake effect, which can hinder the electricity generation potential of downstream wind turbines (Shakoor et al., 2016). However, wind farm projects may also benefit from the established infrastructure of nearby wind farms. The geospatial datasets of the existing wind turbines and wind farm areas were collected from the Electric Energy National Agency (Brasil, 2021c). The datasets were filtered to remove planned and canceled projects, while retaining sites under construction and projects with advanced licensing status, resulting in a compilation of 8246 existing wind turbines (Supplemental Figure S2) primarily concentrated in the northeast and southern states of Brazil.
Power grid
Brazil has an integrated power grid, which could balance electricity distribution between different states within the country (Ferreira et al., 2022). Maintaining a safe distance between transmission lines and wind turbines should be considered in the wind farm site selection. The geospatial data used in this study included the established and planned transmission lines of the integrated national power grid, encompassing voltages ranges from 138 to 750 kV (ONS, 2016).
Calibration of constraint factors
To systematically assess the constraints affecting new onshore wind farm installations, we conducted a statistical analysis based on existing wind turbines in Brazil (Supplemental Table S2). The evaluation focused on two aspects: the mean wind speed at wind turbine locations and the distance between these existing wind turbines and each constraint factor. For each constraint factor, minimum, maximum, and median distance values were calculated using QGIS and compared with values reported in the literature and Brazilian regulation (when applicable). Final buffer values were defined based on this comparative analysis, adopting a conservative approach to support social and environmental sustainability in future wind farm projects. The rationale for each decision is presented in Table 2, and further information is available in the Supplementary Material.
We identify suitable areas for future Brazilian wind farms under two wind power development scenarios, defined using the minimum and median mean wind speed values derived from the statistical analysis (Supplemental Table S2).
High Economic Reward (HER) scenario: This uses the median mean wind speed (8.76 m/s) as the suitability threshold to identify the most promising sites for future wind farm projects, supporting decision-making by local governments and investors.
Wind Power Expansion (WPE) scenario: This uses the minimum mean wind speed observed at existing wind turbine sites (4.49 m/s; Supplemental Table S2) as the threshold to identify minimally suitable areas and estimate the broadest expansion potential for onshore wind energy in Brazil.
Integration of the restrictive map
In this step, a safe distance from each constraint factor were set, according to the equation (1). The suitable area for wind farm projects must keep at least buffer distance away from constraint factor, where is the minimum distance allocated to constraint factor i (Supplemental Table S2).
After generating the restrictive map layer for each Constraint Factor in QGIS, these were then integrated into a final restrictive map layer (equation (2)).
The integrated restrictive map obtained for the WPE scenario (Supplemental Figure S3) demonstrates that approximately 80.41% of Brazil’s land mass, which amounts to 6,847,827 km², is unsuitable for onshore wind farm installation. This is primarily due to the extent of areas with unqualified wind speed, and buffer zones around established infrastructure and Environmental Protection Areas, and other constraint elements. Notably, these unsuitable areas are mainly concentrated in the Northwest region of Brazil.
Determination and integration of evaluation criteria
Determination of evaluation criteria of this study
Evaluation criteria of this study comprise all the technical, economic, and social factors that positively interfere in the suitability of an area for future wind farms. Six evaluation criteria including wind speed, distance to roads, distance to existing wind farms, distance to the power grid, distance to urban settlements and distance to ports and railways are identified based on the existing literature and Brazil’s country profile. Details on the criterion definition and characteristic are presented in the Supplemental Table S3.
Calculation of weights for criteria
MCDM approaches supply diverse options for objectively defining priorities or the weight for each criterion. The Analytic Hierarchy Process (AHP), which was initially designed to reduce decision-making bias by assigning mathematically sound weights to criteria, is among the most widely-used methods (Saaty, 1987). Currently, there are several variations of the AHP method, but those do not necessarily alter the final results (Mosadeghi et al., 2015). Furthermore, the consistency ratio (CR) test within AHP is a safe way to confirm the reliability of the defined values.
The AHP process comprises four phases, beginning with the creation of a pair-wise comparison matrix that assigns relative values when comparing two criteria (Table 3). These values are scored based on a 1–9 scale of relative importance (Supplemental Table S4). The final weight of a criterion is determined by averaging the importance values assigned to it.
The consistency of the matrix can be verified by calculating the CR, where CR is the quotient of the Consistency Index (CI) by the Random Index (RI). A matrix is deemed consistent if the CR < 0.1. After ensuring matrix consistency, the final weights are generated.
To take the different perceptions of relative importance for each criterion, we surveyed seven experts from Brazilian universities and wind power industry, who were selected exclusively if proven nationwide practical and/or academic experience on wind energy in the country (Supplemental Table S5). The surveyed experts were asked to attribute the comparative values based on the complexities and particularities from Brazilian territory. The results of the survey were compiled as shown in Supplemental Table S5. The processed result is used as input to generate the weights for criteria in AHP (Table 3). The CR was checked and confirmed the consistency of the weights.
Wind speed is the most critical factor, as indicated in previous studies. Distance to roads was the second most important criterion, which aligns with Brazil’s heavy reliance on land vehicles for transportation. Proximity to existing wind farms ranked third, as it can reduce infrastructure expenses for new projects. Additionally, proximity to the power grid can reduce the costs of installing transmission lines. Furthermore, shorter distances from future wind farm projects to urbanized areas criteria ranked fifth, while proximity to ports and railways received the lowest weight.
The aggregated weighted criteria map
In this step, we quantified the suitability of each criterion map in QGIS software using equation (3):
Where represents normalized criterion map
Spatial layers of the six evaluation criteria and the wind-speed thresholds applied in the WPE and HER scenarios.
Furthermore, to integrate all the normalized map layers into a single aggregated criteria map in QGIS, the Simple Additive Weighting (SAW) method (Georgiou et al., 2012) was applied (equation (4)).
In equation (4),
With all the criteria maps aggregated, the aggregated weighted criteria maps for the two scenarios were obtained. These two maps illustrate a range of suitability score for onshore wind farm installation shown by the color variation, the value range cannot reach 0 or 1 due to the existence of buffer distance (equation (3)). Supplemental Figures S4a and b represents the WPE and the HER scenarios, respectively.
The generation of the suitability map
After acquiring the integrated restrictive map and the aggregated weighted criteria map, the final suitability maps for onshore wind farm installation in Brazil was generated in QGIS using equation (5) below.
In equation (5),
Data sources
The data collected for the study was compiled from various sources, including Brazilian governmental organizations and established literature (Supplemental Table S6). All geospatial datasets were resampled to a common resolution of 0.1 km × 0.1 km (approximately 100 m) in QGIS prior to integration and analysis. Raster layers with coarser native resolution—notably, mean wind speed data from the Global Wind Atlas (native resolution ~1 km)—were resampled using bilinear interpolation to the 100 m target grid. Vector-format datasets (e.g. urban settlements at 10 m resolution) were rasterized and buffered at the target resolution. The slope layer was retained at 1 km × 1 km due to data availability.
Results
Suitability maps for future wind farms in Brazil
Two national suitability maps were generated for the WPE and HER scenarios, with five categories of suitability that were based on the normalized value intervals (Table 4). After overlapping of the final restrictive map and the final aggregated criteria map (equation (3)), the suitability value ranges from 0.31 to 0.96 under the WPE scenario, and from 0.52 to 0.96 under the HER scenario.
Suitability classes, value intervals, and associated areas under the WPE and HER scenarios.
Figure 3a shows the suitability map for onshore wind farm deployment under the Wind Power Expansion (WPE) scenario (mean wind speed ⩾4.49 m/s). The total suitable area is 1,667,940 km². Of Brazil’s territory, 12.82% and 6.60% are classified as very suitable and extremely suitable, respectively. High-suitability areas are concentrated in the eastern part of the country, with extremely suitable areas predominantly located in the South and Northeast. The dominance of high suitability reflects the co-location of strong wind resources and enabling infrastructure captured by the evaluation criteria. In contrast, low-suitability areas largely coincide with regions excluded by the applied constraint factors.
Suitable areas for future onshore wind farms under the (a) WPE and (b) HER scenarios.
Figure 3b shows suitable areas under the High Economic Reward (HER) scenario (mean wind speed ⩾8.76 m/s). The total suitable area is 14,268 km² (0.17% of Brazil’s continental territory), of which 13,730 km² is classified as very suitable. Suitable areas under HER are mainly concentrated in the Northeastern states of Brazil.
States of Brazil with most suitable areas
Northeastern and southern regions of Brazil comprise the largest proportion of areas suitable for onshore wind farm installation. A zoomed-in view at the suitability map for the HER scenario reveals that the majority of highly suitable areas are concentrated in Bahia (BA) (Figure 4a), Paraiba State (PB) and Rio Grande do Norte (RN) (Figure 4b). These states already have numerous existing wind farms, but there is still a vast space for new onshore wind farm project.
Zoomed-in view of suitable areas under the HER scenario and existing wind farms in (a) Bahia (BA) and (b) Rio Grande do Norte (RN).
To further support planning from local community, government and stakeholders, the list of Brazilian municipalities with the largest shares of extremely suitable areas for onshore wind farm installation are displayed in Table 5, considering the minimum area of 5 km². Specifically, the states of Bahia, Paraiba, and Rio Grande do Norte account for 50.24%, 13.78%, and 12.94% of the total suitable area in the HER scenario, respectively. Supplemental Table S7 lists all Brazilian states with suitable areas for new onshore wind farm installation under HER scenario, and Supplemental Table S8 lists all Brazilian municipalities with extremely suitable available areas under HER scenario.
States and municipalities with the largest suitable areas for wind farms under the HER scenario.
Discussion
Limitations
The assumptions and data availability issues introduce inherent uncertainties and shortcomings in this study. For instance, this study only considers wind speed at a height of 100 m, which is the average hub height of onshore wind turbines in Brazil during the study period. However, wind turbine technology is advancing rapidly (EPE, 2020; Caduff et al., 2012), with newer models reaching hub heights of 150 m or more (von Krauland and Jacobson, 2024). Wind speeds generally increase with height due to reduced surface friction, which could expand the areas suitable for wind farm installation beyond those identified here. Future analyses should incorporate wind data at multiple heights to better capture technological trends and their implications for wind resource assessment. Nevertheless, because 100 m corresponds to the dominant hub height of operating onshore turbines during the study period, this dataset remains appropriate for a national-scale assessment of Brazil’s current wind development stage. Additionally, the resampling process of the wind speed data to a finer grid may convey “false precision.”
It should also be noted that “Distance to Power Grid” was used in this study as a spatial proxy for grid accessibility. However, in operational terms, the feasibility of grid connection depends not only on physical proximity to transmission lines but also—and often more critically—on the available hosting capacity at the point of connection. Transmission congestion is a recognized challenge in Brazil’s northeastern states, where rapid wind energy deployment has in some corridors outpaced grid reinforcement, leading to curtailment events. Publicly available, spatially resolved data on available hosting capacity was not accessible at the national scale required for this study. Future analyses should incorporate grid capacity indicators—such as substation available capacity data from ONS or planned transmission expansion schedules—to better differentiate between “connection proximity” and “connection feasibility.”
The Analytic Hierarchy Process (AHP) was adopted for its transparency, interpretability, and widespread use in renewable energy planning. Nevertheless, as with any expert-based weighting method, AHP introduces a degree of subjectivity that can affect criteria prioritization and spatial outcomes. Although consistency testing indicated reliable expert judgments, reliance on subjective perception remains a limitation compared to data-driven or probabilistic weighting approaches. Future research could increase methodological robustness by integrating fuzzy AHP, entropy weighting, or hybrid multi-criteria techniques that explicitly quantify uncertainty in decision-making (e.g. Ayodele et al., 2018). These approaches would allow for a more nuanced treatment of expert variability and improve the reproducibility of suitability assessments across different geographic or policy contexts.
Regarding the expert survey, although the consulted experts were required to have experience in nationwide renewable energy projects, all Brazilian participants were based in São Paulo or Rio de Janeiro, the country’s largest urban centers. Further studies should consider a larger and more geographically diverse panel of experts with varied professional backgrounds to improve representativeness.
Sensitivity analysis
In multi-criteria decision-making exercises, a “what-if” sensitivity analysis is commonly applied to test the stability of the results against the subjectivity of expert-derived weights (Tegou et al., 2010). In this study, we evaluate how changes in the weights of the six evaluation criteria—wind speed, distance to roads, distance to existing wind farms, distance to the power grid, distance to urban settlements and distance to ports and railways—affect the suitability maps for onshore wind development under both the WPE and HER scenarios. We stress that such changes do not affect the constraint factors, and thus it does not affect the geographic distribution of the suitable areas.
We defined four alternative weighting schemes, plus two sub-scenarios for the wind speed criterion (Supplemental Table S9). The baseline case uses the AHP-derived weights reported in Table 3. Scenario 1 assigns equal weights (0.1667) to all six criteria, representing a neutral configuration where technical, economic, social and environmental dimensions are treated as equally important. Scenario 2 removes the explicitly economic/infrastructure criteria by setting the weights of distance to roads, distance to the power grid and distance to ports and railways to zero and redistributing their weight proportionally among wind speed, distance to existing wind farms and distance to urban settlements. Scenario 3 removes the social criterion by setting the weight of distance to urban settlements to zero and re-normalizing the remaining five criteria. Scenario 4 explores the effect of strengthening or weakening the emphasis on wind resources: in Scenario 4a, the weight of wind speed is increased to 0.6 and the remaining 0.4 is redistributed proportionally across the other criteria; in Scenario 4b, the weight of wind speed is reduced to 0.3 and the remaining 0.7 is redistributed proportionally. For each weighting scheme, we recomputed the suitability index and reclassified it into suitability classes for both WPE and HER (Supplemental Table S10), and we mapped the corresponding spatial patterns (Figures 5 and 6).
Spatial stability of suitability classes under alternative weighting schemes for the WPE scenario.
Spatial stability of suitability classes under alternative weighting schemes for the HER scenario.
For the WPE scenario, the total area classified as suitable (combining all suitability classes) is very similar across all alternative weighting schemes: it decreases only slightly from 1,667,940 km² in the baseline to 1,651,170 km² in Scenarios 1–4 (a difference of about 1%; Supplemental Table S10). In contrast, the internal distribution among suitability classes is highly sensitive to the weights. Under equal weights (Scenario 1), more than four-fifths of the suitable area is classified as “extremely suitable,” while the “very suitable” class shrinks sharply compared with the baseline. When economic/infrastructure criteria are removed (Scenario 2), the share of “extremely suitable” land under WPE drops to less than 1% of the suitable area, whereas “very suitable” land dominates (>85%), reflecting an expansion of high-wind but more remote regions that are no longer penalized by accessibility constraints. Removing the social criterion (Scenario 3) produces more moderate changes, primarily increasing the “very suitable” class at the expense of the “extremely suitable” class, while keeping the overall pattern similar to the baseline. Emphasizing wind speed (Scenario 4a) or, conversely, down-weighting it (Scenario 4b) mainly redistributes area between the “very suitable” and “extremely suitable” classes: Scenario 4b yields a much larger share of land classified as “extremely suitable,” whereas Scenario 4a concentrates most of the suitable area in the “very suitable” class (Supplemental Table S10). Despite these differences in class proportions, the principal clusters of higher suitability under WPE remain concentrated in the Northeast and South regions in all cases (Figure 5).
Under the more restrictive HER scenario, sensitivity patterns are qualitatively similar but affect a much smaller absolute area (around 14,200 km² of suitable land in all weighting schemes; Supplemental Table S10). The total suitable area again varies very little across scenarios (differences below 1%), but the share of land in the highest suitability class is strongly affected by the weights. In the baseline HER map, “extremely suitable” land accounts for about 3% of the suitable area. When wind speed is heavily prioritized (Scenario 4a), this share is reduced to below 1%, as the combination of stricter exclusions and a strong emphasis on wind potential narrows the set of locations that simultaneously satisfy all criteria. In contrast, when wind speed is down-weighted relative to accessibility and other factors (Scenario 4b), the “extremely suitable” class expands to roughly 30% of the suitable area—an almost tenfold increase compared to the baseline. Even in this restrictive scenario, however, the main high-suitability clusters remain concentrated along existing transmission and road corridors in northeastern states such as Rio Grande do Norte, Ceará and Bahia, with additional pockets in the South (Figure 6).
Overall, the sensitivity analysis shows that the GIS–AHP framework is sensitive to the choice of criteria weights with respect to how land is distributed among suitability classes and which specific corridors or municipalities appear most attractive in ranking exercises. At the same time, it confirms that our key strategic conclusion—the Northeast and, to a lesser extent, the South of Brazil consistently emerge as the most promising regions for future onshore wind development—is robust across a broad range of plausible weighting schemes. The sensitivity analysis therefore helps to clarify the trade-offs associated with different decision-making priorities (e.g. maximizing resource quality versus minimizing new infrastructure requirements), while reinforcing the stability of the main national-scale patterns identified in this study.
Spatial and policy implications
The WPE scenario indicates that 80.4% of Brazil’s territory is unsuitable for future wind farms, yet it still leaves over 1.6 million km² available for expansion, thus confirming the country’s vast wind energy potential. This area could theoretically accommodate an additional 15,812 GW of wind power capacity, assuming an installed capacity density of 9.48 MW/km², based on Brazilian wind farms commissioned in 2020 (EPE, 2020; Supplemental Figure S5). This estimate represents 60 times the energy demand expected to be supplied by wind power sources in Brazil by 2050 under the highest electricity-demand scenario. Although simplified and subject to practical barriers (e.g. land ownership complexity, regulatory process, grid integration bottlenecks), its magnitude is enough to confirm Brazil’s potential to become a global reference in wind energy development. The 0.17% of the country’s area categorized as suitable under the HER scenario represents areas that could substantially benefit surrounding urban areas and stimulate regional development, especially if supported by strategic planning.
This study was conceived to serve as a preliminary analysis to support site selection for onshore wind farms. Once the sites are defined, and they must undergo the environmental licensing process mandated by Brazilian law (Brasil, 2014). During this phase, potential positive and negative impacts are analyzed in detail, and detailed mitigation and compensation measures are proposed. However, previous studies have highlighted flaws and corruption in the environmental licensing of past wind farm projects (da Silva and Galvão, 2022; Gorayeb et al., 2018; Nascimento, 2019). By deeming socially and environmentally sensitive areas unsuitable, this study seeks to help prevent potential conflicts during future wind farm licensing. It is also recommended that government authorities strengthen regional environmental agencies and ensure greater transparency throughout the licensing process. For instance, “technical informative meetings” specified in the referred legal document could be made mandatory, as they offer opportunities for local communities to engage with projects, and could help prevent issues related to land tenure security (da Silva and Galvão, 2022; Gorayeb et al., 2018). The opinions of local population should be incorporated into the final decision-making process. Furthermore, this study could also serve as the basis for a transparent, socially inclusive, and responsible territorial zoning process, as suggested in a prior study (Gorayeb et al., 2018), thus providing an updated basis for improved local and regional planning.
To further illustrate the relevance of some of the social constraint factors considered in this study, 0.47% of Brazil’s population consists of Indigenous peoples, encompassing 305 ethnicities living in urban, rural and isolated communities (Cunha, 2022). These communities play an important role in conserving biodiversity and promoting the regrowth of native forests (Alves-Pinto et al., 2022). However, they face continuous pressure from several political and economic interest groups, making it essential to ensure their territorial rights (Baragwanath and Bayi, 2020; Brannstrom et al., 2017). Similar contributions to environmental protection are also observed among the quilombola settlements (Oviedo and Bursztyn, 2016), native communities descended from enslaved Africans who resisted the slavery regime. The catalogued communities of this social group are included in the constraint factors within urbanized settlements.
However, the incorporation of social constraint areas is only first step toward a socially sustainable renewable energy transition in Brazil. Achieving this goal demands a justice-oriented approach. As revealed by the referred studies and thoroughly discussed by Lenoir-Improta and Di Masso (2021), the expansion of wind energy in Brazil often reproduces colonial patterns of development, privileging external investors and marginalizing local populations. It is necessary not only to guarantee public inclusion in energy debates, but also to ground those debates in a collective awareness of the power asymmetries between Global North and South. Furthermore, efforts should focus on establishing governance frameworks for the development of energy infrastructure and policies that ensure a fair distribution of benefits and respect for territorial identities.
From an environmental sustainability perspective, RE projects should adopt a systems approach that extends beyond compliance with the constraint factors considered in this study. Adequate end-of-life logistics for wind turbines begin with material selection and extend through the decommissioning phase of wind farms.
An effectively sustainable project should address the economic, social, and environmental dimensions of sustainability (Goodland, 1995). In this sense, it is worth noting that United Nations Sustainable Development Goals (SDGs) provide a useful framework to operationalize sustainability: RE projects should provide affordable and clean energy (SDG 7), but also contribute to poverty reduction, sustainable cities, climate action, ecosystems support, desertification combat etc. (SDGs 1, 8, 10, 11, 13, 15),(Mungai et al., 2022). Such Goals can be used to support and promote actions in multiple scales of national energy development.
Outlook and transferability
Compared to Archer and Jacobson (2005), which is the only study that evaluated Brazil’s wind potential using coarse reanalysis data at 80 m height, our analysis relies on higher-resolution wind speed data at 100 m and incorporates a broader set of socio-environmental constraints that are individually calibrated for Brazilian national conditions. This allows us to update the spatial picture of onshore wind potential while explicitly accounting for sensitive territories and environmental protection areas.
The resulting suitability maps provide an updated, country-wide assessment that can support national governmental planning for the energy transition, including the definition of preliminary zoning for renewable energy expansion. For instance, a state government that intends to develop a wind energy expansion plan could use our results as a screening tool and then carry out fine-scale studies for the top five municipalities listed in Table 5. The findings can also inform community-level initiatives by offering spatial evidence to support local and regional development strategies aligned with socially inclusive energy growth.
Recent work has begun to explore Artificial Intelligence (AI) and machine learning frameworks for wind farm site selection (Amsharuk et al., 2025). Building on these advances, future research could integrate AI into GIS based–MCDM workflows in several ways. For example, machine learning models could be used to optimize or learn criteria weights from historical project performance data and observed project outcomes, or to predict future land-use and infrastructure changes that may affect site suitability. Deep learning algorithms could also analyze high-resolution satellite imagery to automatically detect and update constraint layers, such as newly urbanized areas or vegetation change, making national suitability assessments more dynamic and responsive to rapidly evolving conditions.
Finally, the method used in this study can be applied to other countries or to smaller regions, particularly in developing economies facing similar socioeconomic and environmental challenges. The model can be adapted with additional and re-calibrated parameters to fit local conditions, and the factors and criteria can be tailored to align with the profile of the new study area, while preserving the overall GIS based–MCDM framework. Also, the methodology of the developed could be updated to integrate temporal wind profiles and energy generation complementarity with other sources of clean energy.
Supplemental Material
sj-docx-1-tss-10.1177_29768632261445736 – Supplemental material for Spatial multi-criteria decision-making for onshore wind farm site selection in Brazil
Supplemental material, sj-docx-1-tss-10.1177_29768632261445736 for Spatial multi-criteria decision-making for onshore wind farm site selection in Brazil by Lucas Garbellini, Shangjun Ke, Srinivasa Raghavendra Bhuvan Gummidi, Di Dong, Morten Birkved, Guotian Cai and Gang Liu in Transactions in Energy and Sustainability
Footnotes
Acknowledgements
This work is financially supported by the National Natural Science Foundation of China (72334001), the Humanities and Social Sciences Fund of the Ministry of Education of China (23JZD018), Independent Research Fund Denmark (ReCAP), and the China Scholarship Council (CSC).
Author contributions
Lucas Garbellini: Data collection, conceptualization, methodology, investigation, formal analysis, writing—original draft. Shangjun Ke: Conceptualization, methodology, investigation, formal analysis, writing—original draft. Srinivasa Raghavendra Bhuvan Gummidi: Conceptualization, methodology, supervision, writing—review and editing. Morten Birkved: Supervision, writing—review and editing. Di Dong: Writing—review and editing. Guotian Cai: Writing—review and editing. Gang Liu: Conceptualization, methodology, supervision, writing—review and editing.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work is financially supported by the National Natural Science Foundation of China (72334001), the Humanities and Social Sciences Fund of the Ministry of Education of China (23JZD018), Independent Research Fund Denmark (ReCAP), and the China Scholarship Council (202006340039).
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data availability statement
Data will be made available on request.
Supplemental material
Supplemental material for this article is available online.
Author biographies
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
